How can I replace a word with its root in postgresql?

Question

what I would like to do is something like

select
"wise, wisdom and wiseness have a common root"  as body,
regexp_replace(body,'wis%','wise')

to obtain

wise --> wise  
wisdom --> wise   
wiseness --> wise

and so the result would be "wise, wise and wise have a common root"

How would you computationally go from "wisdom" to "wise"? It's not that trivial and probably too broad for a StackOverflow question. — Federico klez Culloca, Sep 12 '19 at 12:52
From a linguistic pow I agree, but here it would be enough to go from wis --> to wisdom, which is much easier. I'm not caring about mistakes like 'I wish you're fine' --> 'I wise you're fine' — Edoardo Albergo, Sep 12 '19 at 12:57
Does this need to work with other roots or just specifically with "wis"? — Federico klez Culloca, Sep 12 '19 at 12:58
Just with a specific root. Actually the first comment is exactly what I meant, it was much easier than I thought — Edoardo Albergo, Sep 12 '19 at 13:05

S-Man · Accepted Answer · 2019-09-12T13:05:08.343

4

select
regexp_replace('wise, wisdom and wiseness have a common root','wis[\w]*','wise','g')

RegExp wis[\w]* is searching for all words beginning with wis followed by any number of letters and digits. Use [A-Za-z] instead of [\w] if you only are interested in letters.
Flag g makes it for all occurences.

edited Sep 12 '19 at 13:05

answered Sep 12 '19 at 12:59

S-Man

1 Answers1